Noise suppressing method and device
Technical field
The present invention relates to technical field of image processing, particularly relate to a kind of noise suppressing method and device.
Background technology
Image denoising is a kind of Application comparison technology widely in image procossing, and the object of image denoising is the signal to noise ratio (S/N ratio) in order to improve image, the desired character of outstanding image.
In the shooting process of digital image, noise is present in the links in shooting process, such as image is easily subject to the impact of various factors in the process obtained and transmit, make the image collected by imageing sensor comprise noisy image often, the existence of noise makes the image quality of image to decline to some extent.
Owing to mixing containing noisy noise in image signal and picture signal, make the problems such as image existing characteristics is not obvious, sharpness is low, so usually to need imageing sensor collect image and carry out denoising to improve the signal to noise ratio (S/N ratio) of image, improve the display effect of image.
In prior art; existing multiple method of image being carried out to denoising; for the high-frequency characteristic of the noise in image, usually can use the noise removal function of various low-pass filter realization to image, such as, realize the denoising to image by Gaussian filter, two-sided filter etc.
In the method using low-pass filter denoising, usually need to use certain algorithm template.Treat the pixel of denoising described in containing in described algorithm template, and described in treat in the certain territory of the surrounding of the pixel of denoising pixel.When wave filter works, all pixels in described algorithm template all can participate in computing, calculation of complex, and operand is large.In addition, due to the low-pass characteristic of low-pass filter, the sharp keen image of script on picture can be made to thicken, loss image detail.
The noise of image can be simply divided into brightness noise and color noise usually, brightness noise spatial frequency is higher, and color noise spatial frequency is lower, in order to effectively remove the lower color noise of frequency, need to use the algorithm template larger than size during removal brightness noise, this can make operand during denoising larger, simultaneously due to the employing of large scale algorithm template, the blur effect to image that the characteristic of low-pass filter causes can be more obvious, has a strong impact on the image quality of image.
Summary of the invention
What the present invention solved is when removing color noise to image, easily causes image blurring, loss image detail and the problem that denoising method is complicated, operand is large.
For solving the problem, technical solution of the present invention provides a kind of noise suppressing method, for suppressing the color noise in image; Comprise:
Luma component values based on pixel determines the target zone of carrying out squelch in chromatic component space; The planimetric coordinates space of described chromatic component space for being made up of the first chromatic component and the second chromatic component;
According to the distance between the corresponding position in described chromatic component space of described pixel and described target zone, determine the rejection coefficient of corresponding described pixel;
The first chromatic component value and the second chromatic component value of described pixel is adjusted, to realize the suppression process of the color noise to described pixel by the rejection coefficient of the described pixel of correspondence.
Optionally, described image is yuv format, and described first chromatic component is U component, and described second chromatic component is V component.
Optionally, described image is YCbCr form, and described first chromatic component is Cb component, and described second chromatic component is Cr component.
Optionally, described target zone is any one in border circular areas, elliptical region, square area or rectangular region; The central point of described target zone is the initial point in described planimetric coordinates space.
Optionally, the described luma component values based on pixel determines that the target zone of carrying out squelch in chromatic component space comprises:
When the luma component values of pixel is greater than the first luminance threshold, reduced by the determined described target zone of the luma component values of described pixel along with the increase of the luma component values of pixel;
When the luma component values of pixel is less than the second luminance threshold, reduced by the determined described target zone of the luma component values of described pixel along with the reduction of the luma component values of pixel;
Wherein, described first luminance threshold and the second luminance threshold correspondingly to be determined according to carrying out the noise statistics of described image.
Optionally, described target zone is the border circular areas centered by initial point; The described luma component values based on pixel determines that the target zone of carrying out squelch in chromatic component space comprises:
The radius of described border circular areas is determined by following formula:
Wherein, lum is the luma component values of described pixel, and r is by the radius of the determined described border circular areas of the luma component values of described pixel, lum
0for described second luminance threshold, lum
1for described first luminance threshold, r
maxfor basis is to the radius corresponding to the determined maximum border circular areas carrying out squelch in described chromatic component space of the noise statistics of described image.
Optionally, the span of the rejection coefficient of the described pixel of described correspondence is [0,1].
Optionally, the rejection coefficient of described corresponding pixel points is determined in the following way:
When distance between described pixel and the central point of described target zone is larger, the rejection coefficient of corresponding described pixel is larger; When the central point distance of described pixel and described target zone is less, the rejection coefficient of corresponding described pixel is less.
Optionally, described target zone is the border circular areas centered by initial point; The rejection coefficient of corresponding described pixel is determined by following formula:
Wherein, Ratio is the rejection coefficient of the described pixel of described correspondence, and r is the radius of described border circular areas, and R is the distance value between described pixel and initial point.
Optionally, described image is yuv format; The distance of the central point of described pixel and described target zone passes through formula
determine;
Wherein, R is the distance value between described pixel and described central point, and U is the U component value of described pixel, and V is the V component value of described pixel.
Optionally, the process of the suppression process of the described color noise to described pixel comprises:
By the product of the first chromatic component value of described pixel and the rejection coefficient of the described pixel of described correspondence, as the first chromatic component value of the described pixel suppressed after process;
By the product of the second chromatic component value of described pixel and the rejection coefficient of the described pixel of described correspondence, as the second chromatic component value of the described pixel suppressed after process.
Technical solution of the present invention also provides a kind of Noise Suppression Device, for suppressing the color noise in image; Described device comprises:
Denoising scope determining unit, for determining the target zone of carrying out squelch in chromatic component space based on the luma component values of pixel; The planimetric coordinates space of described chromatic component space for being made up of the first chromatic component and the second chromatic component;
Factor determination unit, for according to the distance between the corresponding position in described chromatic component space of described pixel and described target zone, determines the rejection coefficient of corresponding described pixel;
Suppress processing unit, for being adjusted the first chromatic component value and the second chromatic component value of described pixel by the rejection coefficient of the described pixel of correspondence, to realize the suppression process of the color noise to described pixel.
Compared with prior art, technical scheme of the present invention at least has the following advantages:
Luma component values based on pixel determines the target zone of carrying out squelch in chromatic component space, namely the denoising scope in chrominance space is determined, and then according to the distance determination rejection coefficient between pixel and described denoising scope, adjusted the chromatic component value of described pixel by described rejection coefficient.The method is according to the distribution characteristics of color noise in chromatic component space, determine the corresponding rejection coefficient of corresponding pixel points, and then the suppression process realized described pixel, can effectively remove the color noise in image, computing method are simple, and original picture detail in image can be retained, ensure the sharpness of image frame, improve the image quality of image.
Further, can by the statistics to the noise of image, determine the size of the first luminance threshold and the second luminance threshold accordingly, the target zone of carrying out squelch in chromatic component space can be determined based on described first luminance threshold and the second luminance threshold, can according to the noise statistics of different images, the target zone of the corresponding squelch of adaptive change, can realize effective noise and suppress process, improve the image quality of image.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the noise suppressing method that technical solution of the present invention provides;
Fig. 2 is the schematic flow sheet of the noise suppressing method that the embodiment of the present invention provides;
Fig. 3 is the distribution schematic diagram of color noise in UV space that the embodiment of the present invention provides;
Fig. 4 is the relation schematic diagram of the brightness value of the denoising scope that provides of the embodiment of the present invention and pixel;
Fig. 5 is the schematic diagram that the radius value of the denoising scope that the embodiment of the present invention provides changes accordingly along with the change of the luma component values of pixel;
Fig. 6 is the schematic diagram of the denoising scope of the border circular areas of the different radii value that the embodiment of the present invention provides;
Fig. 7 is the corresponding schematic diagram of the denoising scope that provides of the embodiment of the present invention and rejection coefficient;
Fig. 8 is the structural representation of the Noise Suppression Device that the embodiment of the present invention provides.
Embodiment
In order to improve image image quality, need to suppress picture noise.Usually can by the noise removal function of the realizations such as various low-pass filters to image, but in the process using low-pass filter denoising, need to use certain algorithm template.Because the spatial frequency of the color noise in image is lower, in order to effectively remove the lower color noise of frequency, need to use the algorithm template larger than size during removal brightness noise, this can make operand during denoising larger, simultaneously due to the employing of large scale algorithm template, the blur effect to image that the characteristic of low-pass filter causes can be more obvious, has a strong impact on the image quality of image.
Prior art, when removing color noise to image, exists and easily causes image blurring, loss image detail and the problem that denoising method is complicated, operand is large.
For solving the problem, technical solution of the present invention provides a kind of noise suppressing method, suppresses the color noise in image.
Fig. 1 is the schematic flow sheet of the noise suppressing method that technical solution of the present invention provides, and as shown in Figure 1, first perform step S1, the luma component values based on pixel determines the target zone of carrying out squelch in chromatic component space.
In present specification, in the process that the color noise of image is removed, in order to the color noise of image and brightness noise being made a distinction, image can be transformed in the space containing brightness and chromatic component, such as, image be converted to yuv format, YCbCr form etc.
Find through research, color noise in image, in the coordinate space be made up of the chromatic component of image, within the certain scope centered by true origin of color noise integrated distribution in image in the coordinate space of described chromatic component, and in the coordinate space of described chromatic component, can there is corresponding change along with the change of the luminance component of the pixel of image in the size of the distribution range of the color noise in image.
So the target zone of carrying out squelch in chromatic component space can be determined based on the luma component values of pixel.Described target zone is a region of putting centered by the initial point of described coordinate space, and described target zone can be the region of the shapes such as border circular areas, elliptical region, square area or rectangular region.
Described chromatic component space refers to the planimetric coordinates space be made up of the first chromatic component of image and the second chromatic component, for the image of yuv format, described first chromatic component can be U component, described second chromatic component can be V component, and described chromatic component space refers to the UV coordinate space be made up of U component and V component.For the image of YCbCr form, described first chromatic component can be Cb component, and described second chromatic component can be Cr component.Described chromatic component space refers to the CbCr coordinate space be made up of Cb component and Cr component.
Perform step S2, according to the distance between the corresponding position in described chromatic component space of described pixel and described target zone, determine the rejection coefficient of corresponding described pixel.
For the pixel in image, a position can be determined accordingly in described chromatic component space by the chromatic component of described pixel, and then the distance can determining between the position at described pixel place and described target zone, the rejection coefficient of corresponding described pixel can be determined by described pixel and the distance of the central point (true origin in chromaticity coordinate space) of described target zone.
When the position of pixel is positioned at outside described target zone, illustrate that described pixel is in needs and carries out suppressing, outside the target zone of process, now without the need to carrying out suppression process to pixel, the rejection coefficient of described for correspondence pixel to be set to 1.
When the position of pixel is positioned within described target zone, because color noise is the most concentrated in the center of target zone, so when the central point of described coordinate range, suppression dynamics should be larger, it is smaller that corresponding rejection coefficient can be oppositely arranged, and when away from central point time, suppressing dynamics, corresponding this is smaller, corresponding rejection coefficient can arrange comparatively large a bit.
In present specification, the value of described rejection coefficient can be positioned at [0,1] between, in the position of pixel when the central point of described target zone, the value of described rejection coefficient is more close to zero, when central point away from described target zone of the position of pixel, the value of described rejection coefficient is close to numerical value 1.
Perform step S3, adjusted the first chromatic component value and the second chromatic component value of described pixel by the rejection coefficient of corresponding pixel points, to realize the suppression process of the color noise to described pixel.
After determining the rejection coefficient of corresponding pixel points, just can adjust accordingly the first chromatic component value of described pixel and the second chromatic component value according to described rejection coefficient.Such as, can by the product of the rejection coefficient of the first chromatic component value of pixel and the described pixel of correspondence, as the first chromatic component value of the described pixel suppressed after process; By the product of the second chromatic component value of pixel and the rejection coefficient of the described pixel of correspondence, as the second chromatic component value of the described pixel suppressed after process.Specifically in conjunction with the method that chromatic component value and the luma component values of rejection coefficient to pixel process, do not limit at this.
The noise suppressing method that technical solution of the present invention provides is according to the distribution characteristics of color noise in chromatic component space, determine corresponding rejection coefficient, and then suppression process is carried out to corresponding pixel, can effectively remove the color noise in image, computing method are simple, and original picture detail in image can be retained, ensure the sharpness of image frame, improve the image quality of image.
For enabling above-mentioned purpose of the present invention, feature and advantage more become apparent, and are described in detail specific embodiments of the invention below in conjunction with accompanying drawing.
The maximum impact caused for image due to color noise is, color noise can make originally to present neutral ash, without any the pixel of color, the color that some are light can be shown, the saturation degree of these colors is lower, and real colouring discrimination is comparatively large, but because under the background of neutral ash, very outstanding on picture, visual experience is poor.
In order to suppress color noise, in the present embodiment, the U value and V value needing the pixel carrying out squelch is suppressed in conjunction with rejection coefficient, weakens the color of described pixel, make it draw close as neutral ash gradually.In order to make the color transition nature of image, make to be unlikely to allow the U value of pixel and V value rupture in corresponding space in the process suppressed the color noise of pixel, cause the visual discontinuous of image, need to do gradual change process to the rejection coefficient of corresponding pixel points.In noise range, larger rejection coefficient should be adopted from the pixel away from the central point of described noise range, carry out more weak suppression, and adopt less rejection coefficient from the pixel close to the central point of described noise range, carry out stronger suppression.
Fig. 2 is the schematic flow sheet of the noise suppressing method that the present embodiment provides.To be described the suppression of the color noise in the image of yuv format.
First perform step S201, image is converted to yuv format.
Because the pixel in usual image can be represented by R, G, B tri-color components, in the present embodiment in order to the color noise of pixel and brightness noise are distinguished, first image is transformed into yuv space from rgb space.
The method that image is transformed into yuv space from rgb space can adopt the correlation technique of prior art to carry out, and concrete grammar does not repeat them here.
Perform step S202, the luma component values based on pixel determines the target zone of carrying out squelch in chromatic component space.
Found by research, the color noise in image, in the coordinate space that UV is formed, integrated distribution is within the scope of centered by the initial point of described coordinate space, in the present embodiment, by described noise profile go out as shown in Figure 3 centered by initial point, to connect in rotund region.In coordinate system illustrated in fig. 3, transverse axis represents U component value, and the longitudinal axis represents V component value.Border circular areas centered by initial point represents the distribution range of color noise in UV coordinate space in image.
Due in YUV color space, luminance component Y and chromatic component U, V can represent with different numerical value respectively.In the present embodiment, represent with by digital value 0 to 255 with the span of luminance component Y, the span of chromatic component U and chromatic component V represents to 127 by signed number word value-128.
Because the distribution range of the color noise in image along with the change of the luminance component of the pixel in image, corresponding change can occur.Below in conjunction with Fig. 4, the luma component values of the pixel in image and the distribution relation of color noise are described.As shown in Figure 4, such as, when the luma component values Y of pixel is greater than brightness value Y1, along with the increase of the luma component values of pixel, the luma component values of pixel can more and more close to saturation value 255, and the increase of luma component values along with pixel, the distribution of color noise can more and more near chromatic component origin, the scope that color noise namely can be caused to distribute can reduce gradually, so as shown in Figure 4, when the luma component values of pixel is greater than Y1, along with the increase gradually of the luma component values of pixel, noise range in UV coordinate space can reduce gradually, the target zone (denoising scope) of namely carrying out squelch reduces gradually, when the luma component values Y of pixel is less than brightness value Y0, along with the reduction of the luma component values of pixel, the luma component values of pixel can more and more close to zero, now along with the reduction of the luma component values of pixel, the distribution of color noise equally can more and more near chromatic component origin, namely the distribution range of color noise can be caused equally to reduce gradually, as shown in Figure 4, when the luma component values of pixel is less than Y0, along with the minimizing gradually of the luma component values of pixel, denoising scope in UV coordinate space can reduce gradually, the target zone of namely carrying out squelch reduces gradually.
And when the brightness value of the pixel of image is in described brightness value Y0 and brightness value Y1, the color noise scope in image then can in a metastable noise range, as shown in Figure 4 with r
maxfor in the border circular areas of radius, described with r
maxborder circular areas for radius that is to say that maximum need carry out the scope of squelch.
Can be undertaken by the needs in formula 1 pair of image suppressing the target zone of process to be determined accordingly, that is can determine concrete denoising scope.
Wherein, lum is the luma component values of described pixel, and r is by the radius of the determined described border circular areas of the luma component values of described pixel, lum
0for described second luminance threshold, lum
1for described first luminance threshold, r
maxfor basis is to the radius corresponding to the determined maximum border circular areas carrying out squelch in described chromatic component space of the noise statistics of described image.
Namely Y1 shown in Fig. 4 can be regarded as the first luminance threshold in corresponding formula (1), and namely shown Y0 can be regarded as the second luminance threshold in corresponding formula (1).
When carrying out color noise to image and suppressing process, can by determining described first luminance threshold, the second luminance threshold accordingly to the experimental data of the statistics of the noise data in image and carrying out the radius value r of maximum denoising scope of squelch
max.Can by the brightness range regulating the value of the first luminance threshold and the second luminance threshold to carry out to need to do in adaptive adjustment image the pixel of color noise for different images.
Please refer to Fig. 5, Fig. 5 changes schematic diagram along with the change of the luma component values of pixel accordingly for the radius value r of the denoising scope that the present embodiment provides.When the luma component values of pixel is in the middle of the first luminance threshold and the second luminance threshold, the target zone of carrying out squelch remains unchanged, namely denoising scope as above remains unchanged, and when the luma component values of pixel is greater than the first luminance threshold, described target zone reduces along with the increase of the luma component values of pixel, when the luma component values of pixel is less than the second luminance threshold, described target zone reduces along with the reduction of the luma component values of pixel.The radius value r of described first luminance threshold, the second luminance threshold and maximum denoising scope is obtained in data by experiment
maxafter, for each pixel in image, based on the luma component values of this pixel, the target zone of carrying out squelch corresponding to this pixel can be determined by formula (1).
Perform step S203, calculate pixel and the distance of central point of target zone of carrying out squelch.
Described pixel and the described distance R carrying out the central point of the target zone of squelch can be determined by formula (2).
Wherein, U is the U component value of described pixel, and V is the V component value of described pixel.
In the present embodiment, determine the distance of pixel at UV coordinate space mid-range objectives scope central point by formula (2), in other embodiments, also can determine by other means, concrete mode does not limit.
Perform step S204, determine the rejection coefficient of corresponding described pixel according to the distance between pixel and the central point of described target zone.
Incorporated by reference to reference to figure 6, suppose by step S202 determine by the border circular areas of the determined target zone of carrying out suppressing of the luma component values of described pixel to be radius value as shown in Figure 6 be r0, namely described border circular areas is the denoising scope needing denoising, needs to suppress process accordingly to the pixel within the scope of described border circular areas.
Rejection coefficient corresponding to pixel can carry out corresponding setting according to the distance of the distance between this pixel and the central point of described border circular areas.
Particularly, can be determined the rejection coefficient corresponding to pixel by formula (3).
Wherein, the rejection coefficient of distance value corresponding to the pixel of R of the central point of the border circular areas that it is r0 that Ratio (R) represents with radius value.
Incorporated by reference to reference to figure 7, when the distance value R of the central point of described pixel and described border circular areas is greater than the radius value r0 of border circular areas, illustrating that now pixel is in needs carries out suppressing outside the target zone of process, namely be in outside denoising scope, without the need to carrying out suppression process to it, so the rejection coefficient Ratio (R) corresponding to this pixel can value be 1.And when the distance value R of the central point of described pixel and described border circular areas is less than or equal to r0, illustrate that now pixel is in needs to carry out suppressing in the target zone of process, suppression process need be carried out to it, the rejection coefficient of pixel reduces along with the reduction of described distance value R, accordingly can carry out stronger suppression to described pixel.Particularly, being less than or equal to r0 at described distance value R can adopt the ratio of the R/r0 shown by formula (3) to determine the rejection coefficient of corresponding pixel points.
For different denoising scope, by formula (3) obtain corresponding pixel points rejection coefficient also may be different.
Incorporated by reference to reference to figure 6 and Fig. 7, the denoising scope that radius value is r0 as respectively illustrated in Fig. 6 and radius value are the denoising scope of r1, figure 7 illustrates the different rejection coefficient of the described pixel of correspondence determined for different denoising scopes accordingly.As shown in Figure 7, for with the central point of denoising scope apart from for identical pixel, the rejection coefficient corresponding to denoising scope (the denoising scope as radius value r1) is more among a small circle greater than the rejection coefficient corresponding to denoising scope (the denoising scope as radius value r0) in a big way.
Perform step S205, the U component value of pixel and V component value are multiplied by respectively the rejection coefficient of corresponding described pixel, suppress the U component value after process and V component value as described pixel.
The U component value of described pixel is multiplied by by formula (3) obtain to should the rejection coefficient of pixel, using result of product as the U component value after the suppression process of this pixel, in like manner, the V component value of described pixel is multiplied by described rejection coefficient, using the result of product of correspondence as the V component value after the suppression process of this pixel.
Perform step S206, according to the Y-component value suppressing U, V component value after processing and pixel, obtain the pixel value in described pixel original color space.
After U component value after the suppression process obtaining pixel and V component value, can described pixel be transformed in original color space, such as, in original rgb space, to realize the suppression process to described pixel.
For each pixel in image, the noise suppressing method that all can be provided by the present embodiment, realizes the suppression process to each pixel, and then realizes the suppression to the color noise in image.
The noise suppressing method that the present embodiment provides, adopt the mode that brightness noise is separated with color noise, for the statistical property of color noise, the rejection coefficient of adaptive change corresponding pixel points, realize the suppression process to described pixel, effectively can suppress color noise.
This method can tackle mutually according to the noise statistics data of image determine to carry out to suppress the first luminance threshold in the target zone of process, the second luminance threshold adjusts accordingly, can realize the described target zone of adaptive adjustment, the noisiness for concrete image is carried out effective noise and is suppressed process.
This method effectively can suppress the color noise in image, can retain the picture detail in image, improves the image quality of image, and implementation method is simple, and calculated amount is little.
Corresponding above-mentioned noise suppressing method, the embodiment of the present invention also provides a kind of Noise Suppression Device.Fig. 8 is the structural representation of the Noise Suppression Device that the present embodiment provides.
As shown in Figure 8, described device comprises scope determining unit U11, factor determination unit U12 and suppresses processing unit U13.
Described scope determining unit U11, for determining the target zone of carrying out squelch in chromatic component space based on the luma component values of pixel; The planimetric coordinates space of described chromatic component space for being made up of the first chromatic component and the second chromatic component;
Described factor determination unit U12, for according to the distance between the corresponding position in described chromatic component space of described pixel and described target zone, determines the rejection coefficient of corresponding described pixel;
Described suppression processing unit U13, for being adjusted the first chromatic component value and the second chromatic component value of described pixel by the rejection coefficient of the described pixel of correspondence, to realize the suppression process of the color noise to described pixel.
Described scope determining unit U11 comprises brightness judging unit U111 and the first determining unit U112.
Described brightness judging unit U111, for whether being greater than the first luminance threshold to the luma component values of pixel or whether being less than the second luminance threshold judges.
Described first determining unit U112, for when the luma component values of pixel is greater than the first luminance threshold, determines the increase of the luma component values along with pixel, and the determined described target zone of luma component values of described pixel reduces; When the luma component values of pixel is less than the second luminance threshold constantly, determine that the luma component values along with pixel reduces, the determined described target zone of luma component values of described pixel reduces.
Described factor determination unit U12 comprises metrics calculation unit U121 and the second determining unit U122.
Described metrics calculation unit U121, for calculating the distance of described pixel and described target zone.
Described second determining unit U122, time larger for the distance between described pixel and the central point of described target zone, determines that the rejection coefficient of corresponding described pixel is larger; When distance between described pixel and the central point of described target zone is less, determine that the rejection coefficient of corresponding described pixel is less.
Described suppression processing unit U13 comprises the first colourity processing unit U131 and the second colourity processing unit U132.
Described first colourity processing unit U131, for the product by the first chromatic component value of described pixel and the rejection coefficient of the described pixel of described correspondence, as the first chromatic component value of the described pixel suppressed after process.
Described second colourity processing unit U132, for the product by the second chromatic component value of described pixel and the rejection coefficient of the described pixel of described correspondence, as the second chromatic component value of the described pixel suppressed after process.
Although the present invention discloses as above, the present invention is not defined in this.Any those skilled in the art, without departing from the spirit and scope of the present invention, all can make various changes or modifications, and therefore protection scope of the present invention should be as the criterion with claim limited range.